ISO/IEC 14495-2:2002

Information technology — Lossless and near-lossless compression of continuous-tone still images — Part 2: Extensions

Technologies de l'information — Compression sans perte et quasi sans perte d'images fixes à modelé continu — Partie 2: Extensions

General Information

• Status: Withdrawn
• Publication Date: 03-Apr-2002
• Withdrawal Date: 03-Apr-2002
• ICS: 35.040 - Information coding
• ICS: 35.040.30 - Coding of graphical and photographical information
• Technical Committee: ISO/IEC JTC 1/SC 29 - Coding of audio, picture, multimedia and hypermedia information
• Drafting Committee: ISO/IEC JTC 1/SC 29/WG 1 - JPEG Coding of digital representations of images
• Current Stage: 9599 - Withdrawal of International Standard
• Completion Date: 02-Apr-2003

Relations

• Revised: ISO/IEC 14495-2:2003 - Information technology — Lossless and near-lossless compression of continuous-tone still images: Extensions — Part 2:
• Effective Date: 15-Apr-2008

Standards Content (Sample)

INTERNATIONAL ISO/IEC
STANDARD 14495-2
First edition
2002-03-15
Information technology — Lossless and
near-lossless compression of
continuous-tone still images —
Part 2:
Extensions
Technologies de l'information — Compression sans perte et quasi sans
perte d’images fixes à modelé continu —
Partie 2: Extensions
Reference number
ISO/IEC 14495-2:2002(E)
©
ISO/IEC 2002

---------------------- Page: 1 ----------------------
ISO/IEC 14495-2:2002(E)
PDF disclaimer
This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but shall not
be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In downloading this
file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat accepts no liability in this
area.
Adobe is a trademark of Adobe Systems Incorporated.
Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation parameters
were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In the unlikely event
that a problem relating to it is found, please inform the Central Secretariat at the address given below.
© ISO/IEC 2002
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means, electronic
or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or ISO's member body
in the country of the requester.
ISO copyright office
Case postale 56 � CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail [e-mail]
Web www.iso.ch
Printed in Switzerland
ii © ISO/IEC 2002 – All rights reserved

---------------------- Page: 2 ----------------------
ISO/IEC 14495-2:2002(E)
Contents
1 Scope. 1
2 Normative references . 1
3. Definitions, abbreviations, symbols and conventions . 2
4. General . 4
5. Interchange format requirements. 6
6. Encoder requirements. 6
7. Decoder requirements. 7
8. Conformance testing for extensions . 7
Annex A Encoding procedures with arithmetic coding for a single component. 10
A.1 Coding parameters and compressed image data. 10
A.2 Initialisations and conventions . 10
A.3 Context determination. 12
A.4 Prediction . 14
A.5 Prediction error encoding. 16
A.6 Update variables. 20
A.7 Flow of encoding procedures . 22
Annex B Arithmetic coding. 24
B.1 Arithmetic encoding procedures . 24
B.2 Arithmetic decoding procedures . 28
Annex C Encoding with arithmetic coding for multiple component images. 30
C.1 Introduction. 30
C.2 Line interleaved mode. 30
C.3 Sample interleaved mode . 30
C.4 Minimum coded unit (MCU) . 31
Annex D Extended functions for the baseline coding model. 32
D.1 Extensions of near-lossless coding. 32
D.2 Extensions of prediction on baseline coding model . 33
D.3 Extension of Golomb coding. 36
Annex E Fixed length coding. 38
E.1 Introduction. 38
E.2 Fixed length coding. 38
Annex F Sample transformation for inverse colour transform. 39
F.1 Inverse colour transform . 39
F.2 Example and guideline (Informative). 40
Annex G Compressed data format. 43
G.1 General aspects of the compressed data format specification . 43
Annex H Control procedures for extensions. 50
H.1 Control procedure for encoding a restart interval. 50
H.2 Control procedure for encoding a minimum coded unit (MCU) with fixed length code (FLC) . 50
Annex I Conformance tests. 53
I.1 Test images . 53
Annex J Patents . 55
Annex K Bibliography. 57
© ISO/IEC 2002 – All rights reserved iii

---------------------- Page: 3 ----------------------
ISO/IEC 14495-2:2002(E)
Foreword
ISO (the International Organization for Standardization) and IEC (the International Electrotechnical Commission) form the
specialized system for worldwide standardization. National bodies that are members of ISO or IEC participate in the
development of International Standards through technical committees established by the respective organization to deal with
particular fields of technical activity. ISO and IEC technical committees collaborate in fields of mutual interest. Other
international organizations, governmental and non-governmental, in liaison with ISO and IEC, also take part in the work.
International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 3.
In the field of information technology, ISO and IEC have established a joint technical committee, ISO/IEC JTC 1. Draft
International Standards adopted by the joint technical committee are circulated to national bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the national bodies casting a vote.
International Standard ISO/IEC 14495-2 was prepared by Joint Technical Committee ISO/IEC JTC 1, Information technology,
Subcommittee SC 29, Coding of audio, picture, multimedia and hypermedia information, in collaboration with ITU-T, but is
not published as common text at this time.
ISO/IEC 14495 consists of the following parts, under the general title Information technology — Lossless and near-lossless
compression of continuous-tone still images:
� Part 1: Baseline
� Part 2: Extensions
Annexes A to I form a normative part of this part of ISO/IEC 14495. Annexes J and K are for information only.
iv © ISO/IEC 2002 – All rights reserved

---------------------- Page: 4 ----------------------
INTERNATIONAL STANDARD ISO/IEC 14495-2:2002(E)
INFORMATION TECHNOLOGY —
LOSSLESS AND NEAR-LOSSLESS COMPRESSION OF CONTINUOUS-TONE
STILL IMAGES — PART 2: EXTENSIONS
1Scope
This Recommendation | International Standard defines a set of lossless (bit-preserving) and nearly lossless (where the error for
each reconstructed sample is bounded by a pre-defined value) compression methods for coding continuous-tone (including bi-
level), gray-scale, or colour digital still images.
This Recommendation | International Standard
– specifies extensions (including arithmetic coding, extension of near lossless coding, extension of prediction and
extension of Golomb coding) to processes for converting source image data to compressed image data
– specifies extensions to processes for converting compressed image data to reconstructed image data including
an extension for sample tranformation for inverse colour transforms
– specifies coded representations for compressed image data
– provides guidance on how to implement these processes in practice
2 Normative references
The following Recommendations and International Standards contain provisions which, through references in this text,
constitute provisions of this Recommendation | International Standard. At the time of publication, the editions indicated were
valid. All Recommendations and International Standards are subject to revision, and parties to agreements based on this
Recommendation | International Standard are encouraged to investigate the possibility of applying the most recent editions of
the Recommendations and International Standards listed below. Members of IEC and ISO maintain registers of currently valid
International Standards. The ITU-T Telecommunication Standardization Bureau (TSB) maintains a list of the currently valid
ITU-T Recommendations.
2.1 Identical Recommendations | International Standards
– CCITT Recommendation T.81 (1992) � ISO/IEC 10918-1:1994, Information technology – Digital compression
and coding of continuous-tone still images: Requirements and guidelines.
– ITU-T Recommendation T.83 (1994) � ISO/IEC 10918-2:1995, Information technology – Digital compression
and coding of continuous-tone still images: Compliance testing.
– ITU-T Recommendation T.84 (1996) � ISO/IEC 10918-3:1997, Information technology – Digital compression
and coding of continuous-tone still images: Extensions.
– ITU-T Recommendation T.87 (2000) � ISO/IEC 14495-1:2000, Information technology – Lossless and near-
lossless compression of continuous-tone still images: Baseline.
2.2 Additional references
- ISO/IEC 646:1991, Information technology - ISO 7-bit coded character set for information interchange.
- ISO 5807:1985, Information processing - Documentation symbols and conventions for data, program and system
flowcharts, program network charts and system resources charts.
- ISO/IEC 9899:1999, Programming languages - C.
© ISO/IEC 2002 – All rights reserved 1

---------------------- Page: 5 ----------------------
ISO/IEC 14495-2:2002(E)
3. Definitions, abbreviations, symbols and conventions
3.1 Definitions
For the purposes of this Recommendation | International Standard, the following definitions apply in addition to the definitions
used in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000.
3.1.1 arithmetic encoder: An embodiment of an arithmetic encoding procedure.
3.1.2 arithmetic encoding: A procedure which encodes a sample as a binary representation of the sequence of previously
encoded samples by means of a recursive subdivision of a unit interval.
3.1.3 arithmetic decoder: An embodiment of an arithmetic decoding procedure.
3.1.4 arithmetic decoding: A procedure which recovers source data from an encoded bit stream produced by an arithmetic
encoder.
3.1.5 binary context: Context used to determine the binary arithmetic coding of the present binary decision
3.1.6 binary decision: Choice between two alternatives.
3.1.7 colour transform: A procedure for sample transformation for inverse colour transform.
3.1.8 sign flipping: The procedure which reverses the sign of a prediction error according to accumulated prediction
errors.
3.1.9 symbol packing: A procedure which may be applied to source images in which sample values are sparsely
distributed.
3.1.10 visual quantization: An extended function of near-lossless coding which enables to change the difference bound
according to the context.
3.2 Abbreviations
In additions to the abbreviations used in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000, the abbreviations used in this
Recommendation | International Standard are listed below.
FLC: Fixed length code.
LPS: Less probable symbol
MPS: More probable symbol
3.3 Symbols
In addition to the symbols used in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000, the symbols used in this Recommendation |
International Standard are listed below. A convention is used that parameters which are fixed in value during the encoding of a
scan are indicated in boldface capital letters, and variables which change in value during the encoding of a scan are indicated
in italicised letters.
Areg current numerical-line interval being renormalized
ArithmeticEncode() a function in the C programming language
Av[0.30] 31 constants corresponding to LPS probability estimate
Avd auxiliary variable storing modified Av
BASIC_T1, BASIC_T2, BASIC_T3, BASIC_T4 basic default threshold values
Bin binary decision
Buf[0.1] bytes stored to avoid carry-over propagation to the encoded bit stream
Creg value of code register storing the trailing bits of the encoded bit stream
ENT indication of the coding process used for the scan
2 © ISO/IEC 2002 – All rights reserved

---------------------- Page: 6 ----------------------
ISO/IEC 14495-2:2002(E)
Flag[0.MAXVAL] MAXVAL+1 flags which indicate if corresponding sample values already occurred
GetBinaryContext() a function in the C programming language
GetByte() a function in the C programming language
GetGolombk() a function in the C programming language
Hd auxiliary variable storing an integer value corresponding to a half of the full range but shifted according to the
size of the current interval
LPScnt[0.MAXS] accumulated occurrence count of the LPS (less probable symbol) at each binary context
MAXcnt threshold value at which MLcnt and LPScnt are halved
MAXS maximum index of binary contexts
MLcnt[0.MAXS] accumulated occurrence count of each binary context.
MPSvalue[0.MAXS] sense of the MPS (more probable symbol) at each binary context
nearq context-dependent difference bound for near-lossless coding using visual quantization.
NEARRUN difference bound for near-lossless coding in run mode.
NMCU number of MCU’s
Prob LPS probability estimated from MLcnt and LPScnt
Qx the (quantized) value of a sample to be encoded with fixed length code
S index for binary contexts
SOF JPEG-LS frame marker for this extension
57
SPf[0.RANGE] RANGE+1 flags indicating if corresponding mapped error values already occurred
SPm[0.RANGE] mapping table of MErrval or EMErrval for symbol packing
SPt the smallest positive integer greater than all mapped error values that occurred in the scan up to this point.
SPx number of the different mapped error values that already occurred.
T1, T2, T3 thresholds for local gradients
T4 threshold for an additional local gradient
TEMErrval auxiliary variable storing EMErrval.
Th[0.29] threshold to determine suitable value of Av
TMErrval auxiliary variable storing MErrval
TQ visual quantization threshold
wct number of bits by which Areg is shifted
Zerograd flag indicating local gradients are all zero.
4. General
The purpose of this clause is to give an overview of this Recommendation | International Standard.
This Recommendation | International Standard defines extensions to the elements specified in ITU-T Rec. T.87 | ISO/IEC
14495-1:2000. Extensions which pertain to encoding or decoding are defined as procedures which may be used in
combination with the encoding and decoding processes of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000. This Recommendation |
International Standard also defines extensions to the compressed data formats, i.e., interchange. Each encoding or decoding
extension shall only be used in combination with particular coding processes and only in accordance with the requirements set
© ISO/IEC 2002 – All rights reserved 3

---------------------- Page: 7 ----------------------
ISO/IEC 14495-2:2002(E)
forth herein. These extensions are backward compatible in the sense that decoders which implement these extensions will also
support configuration subsets that are currently defined by ITU-T Rec. T.87 | ISO/IEC 14495-1:2000.
4.1 Extensions specified by this Recommendation | International Standard
The following extensions are specified:
— An extension which provides for arithmetic coding. This extension will provide higher compression ratio,
especially with high-skew images.
— An extension which provides for variable near-lossless coding. This extension will provide a wider variety of
possible nearly lossless reconstructions of a source image than ITU-T Rec. T.87 | ISO/IEC 14495-1:2000. There
are two types of variable near-lossless coding, depending on whether the difference bound is changed:
a) according to its context; or
b) in vertical direction.
— An extension which provides for a modified prediction procedure in source images in which sample values are
sparsely distributed.
— An extension which provides for a modified Golomb coding procedure. This modification avoids possible
expansion of compressed image data, and improves coding efficiency by eliminating code words that are not
used.
— An extension which provides for fixed length coding.
— An extension which provides for a modified sample transformation process. This extension can be used to
define inverse colour transforms in order to achieve greater efficiency by compressing a source image in a
different colour representation.
The following sections describe these extensions in greater detail.
4.1.1 Encoding with arithmetic coding
In JPEG-LS baseline, specified in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000, simple but efficient coding is achieved by the
combination of Golomb coding (regular mode) and the run mode. However, for some very high-skewed image data such as
computer generated images, the compression efficiency is affected by the use of symbol-by-symbol coding in contexts that
present highly skewed distributions. Therefore, coding procedures based on arithmetic coding are specified in this
Recommendation | International Standard as an extended function, which enables alphabet extension for every context (rather
than only in run mode) and provides higher compression performance with a moderate increase of the coder complexity.
The arithmetic coder adopted in this Recommendation | International Standard is characterized by its fast multiplication-free
arithmetic operation and radix-255 representation. The details are described in Annex A and B.
4.1.2 Extension of Near-lossless coding
The extension of the near-lossless coding capabilities of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 is to allow the NEAR
parameter to vary during the process of encoding a source image. There are two types of variable near-lossless coding, serving
two different purposes.
4.1.2.1 Visual quantization
Visual quantization takes into account the human visual system by primarily performing the quantization in high activity
regions of the image where the activity may mask for the quantization noise. Therefore, by extending the near-lossless coding
capabilities of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 so that the NEAR parameter can change according to its context, it
becomes possible to provide reconstructed images whose distance from the source image is between those obtained with
compression schemes using NEAR=n and NEAR=n+1, where n denotes a non-negative integer.
4.1.2.2 Re-specification of NEAR value
This type of variable near-lossless coding can vary the NEAR parameter according to the vertical direction. The main purpose
of this extension of the near-lossless coding capabilities of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 is to provide a
mechanism by which an encoder can change the value of NEAR according to the observed compressibility of the source
image, which is useful to control the total length of compressed image data within some specified amount. By this extension,
an encoder can compress a source image to less than a pre-specified size with a single sequential pass over the image. The
capability is valuable to applications which utilize a fixed-size compressed image memory.
4 © ISO/IEC 2002 – All rights reserved

---------------------- Page: 8 ----------------------
ISO/IEC 14495-2:2002(E)
4.1.3 Extension of prediction
The prediction and error coding procedure specified in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 is not suitable for images
with sparse histograms, such as limited color images or fewer-bit images expressed by byte form. These images contain only a
subset of the possible sample values in each component, and the edge-detecting predictor specified in Figure A.5 of ITU-T
Rec. T.87 | ISO/IEC 14495-1:2000 would tend to concentrate the value of the prediction errors into a reduced set. However,
the prediction correction procedure specified in Figure A.6 of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 tends to spread these
values over the entire range. In addition, the Golomb coding procedure would assign short code words to small prediction
errors, even those that do not occur in the image component.
The goal of this extension is to modify the prediction procedure in order to alleviate this unwanted effect by checking the
occurrence of the corrected predicted value Px in the past. This extension also provides a modified coding procedure to
improve the coding efficiency for these images.
4.1.4 Extension of Golomb coding
In addition to the specification of Golomb coding defined in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000, two modifications are
incorporated in this Recommendation | International Standard as follows:
4.1.4.1 Golomb code completion
More effective usage of Golomb code words is specified in this Recommendation | International Standard, in which the final
bit “1” in the longest possible unary representation used in a Golomb code, which is redundant, shall be omitted. This
procedure improves the coding efficiency especially in cases in which this Recommendation | International Standard is also
applied to bi-level images.
4.1.4.2 Omission of run interruption sample coding
In cases in which this Recommendation | International Standard is also applied to bi-level images and the mode is not sample
interleaved, the encoding of the run interruption sample is superfluous and shall be omitted.
4.1.5 Fixed length coding
In this Recommendation | International Standard, a procedure to avoid situations in which the compressed image data is larger
than the source image data is incorporated. The values of image samples, or the quantized values in near-lossless coding, are
encoded with a fixed length code. The region in a scan to be encoded with a fixed length code is specified by appending a
marker indicating the beginning of the fixed length coding and the end of the fixed length coding after an appropriate MCU.
4.1.6 Sample transformation for inverse colour transforms
In this Recommendation | International Standard, a procedure for sample transformation is provided, in addition to the ones
defined in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000. This procedure uses corresponding values of decoded samples in the
various components, to reconstruct the source image data, which is of the same precision as the encoded data. The goal of this
extension is to facilitate the use of this Recommendation | International Standard in conjunction with colour transforms to
improve coding efficiency.
4.2 Descriptions of extended functions
The coding procedure specified in Annex A of ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 is referred to as baseline coding
process. The newly introduced coding procedure, modified from the baseline coding process, is referred to as arithmetic-based
process. The context modelling for the arithmetic-based coding process is described in Annex A, and the arithmetic coding
procedure of a binary symbol for the given context is described in Annex B. The functions outlined in 4.1.2, 4.1.3,and 4.1.6,
can be used on either arithmetic-based coding process or baseline coding process. The functions outlined in 4.1.4 and 4.1.5.
can be used only in baseline coding process. All the extended functions are optional and the combinations of the extended
functions are arbitrary under this general rule.
The use of the extended functions outlined in 4.1.2 and 4.1.3 in conjunction with the arithmetic-based coding process is also
described in Annex A. The use of the extended functions outlined in 4.1.2 and 4.1.3 in conjunction with the baseline coding
process is described in Annex D by referring to the differences with respect to the coding process of the non-extended
functions described in ITU-T Rec. T.87 | ISO/IEC 14495-1:2000.The functions outlined in 4.1.4 are also described in Annex
D.
The extended functions outlined in 4.1.5 are described in Annex E. The extended functions outlined in 4.1.6 on both
arithmetic-based and baseline coding are described in Annex F.
© ISO/IEC 2002 – All rights reserved 5

---------------------- Page: 9 ----------------------
ISO/IEC 14495-2:2002(E)
The contents of the Annexes for extended functions are summarized in Table 1.
Table 1 – Combination of extended functions and corresponding Annex
Extension of Extension of Arithmetic Extension of Fixed length Colour
near-lossless prediction coding Golomb coding transform
coding procedure coding
Baseline coding ---
Annex D Annex D Annex D Annex E
process
Annex F
Arithmetic-based --- ---
Annex A Annex A Annex B
coding process
5. Interchange format requirements
The interchange format is the coded representation of compressed image data for exchange between application environments.
The interchange format requirements are that any compressed image data represented in interchange format shall comply with
the syntax and code assignments appropriate for the coding process and extensions selected, as defined in Annex C of ITU-T
T. Rec. 87 | ISO/IEC 14495-1:2000 and Annex G of this Recommendation | International Standard.
6. Encoder requirements
An encoding process converts source image data to compressed image data. ITU-T Rec. T.87 | ISO/IEC 14495-1:2000
specifies baseline coding processes. This Recommendation | International Standard defines arithmetic-based coding process
and encoding extensions which may be used in combination with baseline coding process or arithmetic-based coding process.
An extended encoder is an embodiment of one (or more) of the encoding processes specified in this Recommendation |
International Standard or ITU-T Rec. T.87 | ISO/IEC 14495-1:2000 used in combination with one or more of the encoding
extensions specified herein. In order to comply with this Recommendation | International Standard, an extended encoder shall
satisfy at least one of the following two requirements.
An extended encoder shall
a) convert source image data to compressed image data which conform to the interchange format syntax specified
in Annex G;
b) convert source image data to compressed image data which comply with the abbreviated format for compressed
image data syntax specified in Annex G.
Conformance tests for the above requirements are specified in clause 8 of this Recommendation | International Standard.
NOTE – There is no requirement in this Recommendation | International Standard that any encoder which embodies one of the
encoding processes and extensions shall be able to operate for all ranges of the parameters which are allowed. An encoder is only required to
meet the applicable conformance tests, and to generate th
...